Scholarlink

<p>Breast cancer is a leading cause of mortality among women worldwide, primarily due to abnormal cell activity in the breast. Early detection and accurate classification are vital for effective treatment. While traditional diagnostic methods such as surgical biopsy, fine needle aspiration, and mammography are commonly used, they often have limited sensitivity and effectiveness. To overcome these limitations, an optimized You Only Look Once Version 3 (YOLOv3) technique is employed to analyze ultrasound images, extract key features, and classify breast tumors as benign, malignant, or normal. Increasing the number of training datasets improves the network's accuracy, reduces errors, and enhances the overall system performance in distinguishing between normal, benign, and cancerous tissues. The primary dataset used for training and testing is the Wisconsin Breast Cancer Dataset (WBCD) ultrasound images, along with a few additional datasets. The optimized YOLOv3 achieves accuracy, sensitivity, and specificity of 97.21%, 93.63%, and 96.81%, respectively. This approach outperforms existing techniques, providing a reliable computer-assisted diagnostic system to aid radiologists in clinical settings.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning technique for breast cancer detection and classification

  • 期刊论文

  • 发表日期：

    2025年4月4日

查看全文
  • N. Kavitha,
  • P. Madhumathy,
  • R. Manjunatha Prasad,
  • D. N. Chandrappa

摘要

Breast cancer is a leading cause of mortality among women worldwide, primarily due to abnormal cell activity in the breast. Early detection and accurate classification are vital for effective treatment. While traditional diagnostic methods such as surgical biopsy, fine needle aspiration, and mammography are commonly used, they often have limited sensitivity and effectiveness. To overcome these limitations, an optimized You Only Look Once Version 3 (YOLOv3) technique is employed to analyze ultrasound images, extract key features, and classify breast tumors as benign, malignant, or normal. Increasing the number of training datasets improves the network's accuracy, reduces errors, and enhances the overall system performance in distinguishing between normal, benign, and cancerous tissues. The primary dataset used for training and testing is the Wisconsin Breast Cancer Dataset (WBCD) ultrasound images, along with a few additional datasets. The optimized YOLOv3 achieves accuracy, sensitivity, and specificity of 97.21%, 93.63%, and 96.81%, respectively. This approach outperforms existing techniques, providing a reliable computer-assisted diagnostic system to aid radiologists in clinical settings.